Study On The Fault Diagnosis Method For Transformer And Circuit Breaker In Power System

With the rapid development of the national economy,the demand for electrical energy in various industries has gradually increased.The stable supply of electrical energy plays an important role in the development of various industries,accompanied by the promotion of policies such as “smart grid” and “Made in China 2025”,for the power system.Stability and intelligence also raise new requirements.In power systems,power transformers and high voltage circuit breakers are widely used,and they are indispensable for two important and frequent failures.Failure of transformers and high voltage circuit breakers can cause great harm and induce major power accidents.Completed with the support of the Provincial Natural Science Foundation of China,the traditional fault diagnosis methods can no longer meet the requirements of fast,accurate,intelligent,stable and existing fault diagnosis methods.It proposes a fusion artificial intelligence method for transformers and high voltage circuit breakers.Fault diagnosis methods have important research value and practical application prospects.This paper analyzes the physical structure,working principle,and common faults of transformers and high-voltage circuit breakers in power systems,and introduces common fault types and fault characterizations of transformers and high-voltage circuit breakers.An improved K-means clustering algorithm based on particle swarm optimization is proposed.The proposed method improves the clustering center initialization mode of the traditional K-means algorithm,and improves the distance evaluation function.It adds a weighting factor to each dimension of the input vector to construct a new distance evaluation function;then directly uses the percentage of gas in transformer oil as the The fault diagnosis data avoids the loss of information caused by the ratio method in the encoding process and achieves a higher accuracy of the diagnostic effect.This paper presents a multi-classification relevance vector machine fault diagnosis method.This method is aimed at high-voltage circuit breakers with complex structure and special working conditions.The faults have more categories and are not easily able to acquire a large number of fault sample data.The relevance vector machine method with good classification effect under small sample data conditions The combination of the "1-to-1" multi-classification model with higher accuracy makes it possible to perform both multi-fault classification and high accuracy.The experimental simulation results show that the K-means algorithm based on PSO proposed in this paper can obtain accurate results in the field of transformer fault diagnosis;the proposed multi-classification relevance vector machine algorithm can have a smaller BP neural network algorithm on small data sets.And the support vector machine algorithm has better classification accuracy and classification speed,and has higher diagnostic credibility on the small fault sample data set of the high voltage circuit breaker.